{
  "id": 248207,
  "title": "Should you let Kaggle choose the best submissions?",
  "url": "/competitions/coleridgeinitiative-show-us-the-data/discussion/248207",
  "author_name": "Rajat Gupta",
  "post_date": "2021-06-22T18:49:00.993000",
  "votes": 1,
  "comment_count": 5,
  "views": 0,
  "content": "<p>I have a couple of scores as 0.576. However, we don't know the score precisely after the 3rd decimal point. So, should I let Kaggle choose my two best submissions from the public leaderboard as I wouldn't know which two are the best among 0.576s?</p>",
  "messages": [
    {
      "id": 1361362,
      "postDate": "2021-06-22T18:49:00.993Z",
      "content": "<p>I have a couple of scores as 0.576. However, we don't know the score precisely after the 3rd decimal point. So, should I let Kaggle choose my two best submissions from the public leaderboard as I wouldn't know which two are the best among 0.576s?</p>",
      "rawMarkdown": "I have a couple of scores as 0.576. However, we don't know the score precisely after the 3rd decimal point. So, should I let Kaggle choose my two best submissions from the public leaderboard as I wouldn't know which two are the best among 0.576s?",
      "votes": 1
    },
    {
      "id": 1361378,
      "postDate": "2021-06-22T19:03:42.340Z",
      "content": "<p>In this competition, definitely no!! Also, remember your best public score may not work well on private. Don't get blinded by public lb. You'll have to make good decision on what can generalise well on private. </p>",
      "rawMarkdown": "In this competition, definitely no!! Also, remember your best public score may not work well on private. Don't get blinded by public lb. You'll have to make good decision on what can generalise well on private. ",
      "votes": 2,
      "replies": [
        {
          "id": 1361386,
          "postDate": "2021-06-22T19:12:47.203Z",
          "content": "<p>I am not sure if the public leaderboard score is even representative of the private leaderboard score since the public score is only on 12% of the data. How should I choose my scores then?</p>",
          "rawMarkdown": "I am not sure if the public leaderboard score is even representative of the private leaderboard score since the public score is only on 12% of the data. How should I choose my scores then?"
        },
        {
          "id": 1361480,
          "postDate": "2021-06-22T21:07:50.910Z",
          "content": "<p>That would depend on your approach.</p>",
          "rawMarkdown": "That would depend on your approach."
        }
      ]
    },
    {
      "id": 1361761,
      "postDate": "2021-06-23T03:56:00.523Z",
      "content": "<p>Well, my best scores among the 0.576s chosen by Kaggle actually turned out to the best on private LB as well. Sometimes best public LB scores translate to best private LB scores too.</p>",
      "rawMarkdown": "Well, my best scores among the 0.576s chosen by Kaggle actually turned out to the best on private LB as well. Sometimes best public LB scores translate to best private LB scores too."
    },
    {
      "id": 1361385,
      "postDate": "2021-06-22T19:11:08.950Z",
      "rawMarkdown": "",
      "isDeleted": true
    }
  ],
  "comments": [
    {
      "id": 1361378,
      "author_name": "Kumar Shubham",
      "author_url": "",
      "post_date": "2021-06-22T19:03:42.340000",
      "content": "<p>In this competition, definitely no!! Also, remember your best public score may not work well on private. Don't get blinded by public lb. You'll have to make good decision on what can generalise well on private. </p>",
      "votes": 2,
      "replies": [
        {
          "id": 1361386,
          "author_name": "Rajat Gupta",
          "author_url": "",
          "post_date": "2021-06-22T19:12:47.203000",
          "content": "<p>I am not sure if the public leaderboard score is even representative of the private leaderboard score since the public score is only on 12% of the data. How should I choose my scores then?</p>",
          "votes": 0,
          "replies": []
        },
        {
          "id": 1361480,
          "author_name": "Pablo Carbó",
          "author_url": "",
          "post_date": "2021-06-22T21:07:50.910000",
          "content": "<p>That would depend on your approach.</p>",
          "votes": 0,
          "replies": []
        }
      ]
    },
    {
      "id": 1361761,
      "author_name": "Rajat Gupta",
      "author_url": "",
      "post_date": "2021-06-23T03:56:00.523000",
      "content": "<p>Well, my best scores among the 0.576s chosen by Kaggle actually turned out to the best on private LB as well. Sometimes best public LB scores translate to best private LB scores too.</p>",
      "votes": 0,
      "replies": []
    },
    {
      "id": 1361385,
      "author_name": "",
      "author_url": "",
      "post_date": "2021-06-22T19:11:08.950000",
      "content": "",
      "votes": 0,
      "replies": []
    }
  ],
  "raw_markdown_by_id": {
    "1361362": "I have a couple of scores as 0.576. However, we don't know the score precisely after the 3rd decimal point. So, should I let Kaggle choose my two best submissions from the public leaderboard as I wouldn't know which two are the best among 0.576s?",
    "1361378": "In this competition, definitely no!! Also, remember your best public score may not work well on private. Don't get blinded by public lb. You'll have to make good decision on what can generalise well on private. ",
    "1361761": "Well, my best scores among the 0.576s chosen by Kaggle actually turned out to the best on private LB as well. Sometimes best public LB scores translate to best private LB scores too.",
    "1361385": ""
  }
}